I recently bought PureRAW 2 to batch-process my images with DeepPRIME and noticed it felt pretty slow. In the past I used PhotoLab 4 to process my images but it felt even slower than that.
I looked at Task Manager and noticed that neither my CPU nor GPU were anywhere close to fully utilized. Most of the time, the computer was barely doing anything.
I wanted to see if PhotoLab 4/5 really were faster and/or making better use of my hardware, so I took a set of 60MP ARW files from my Sony A7RIV and processed them with both pieces of software, recording my computer’s CPU/GPU load at the same time.
Here are the resulting charts:
You can see that the orange line for PureRAW 2 is close to zero around 2/3rds of the time, this is because it’s not touching the GPU at all. PhotoLab 4 and 5 on the other hand both keep the GPU occupied at nearly all times.
The practical result was that I was right: PureRAW 2 is much slower than PhotoLab, both 4 and 5:
|Time per image
My PhotoLab results are pretty close to others on the DeepPRIME benchmark spreadsheet, PureRAW 2 seems to be the slow oddity.
So in my opinion, avoid PureRAW 2, buy PhotoLab 5 instead.
I’m working through this with support and I’ll update here if they offer any fixes.
Have you tried activating your GPU directly in Preferences, instead of auto?
Yes, these charts have the GPU explicitly selected.
Yes that should work in PR2 much faster than 43s. I hope that support can help you out.
FWIW, below is some data from timing PureRAW 2 (v 2.0.1 build 1) on an M1 Max MacBook Pro with 30 Sony A1 ARW files (49.768Mpix).
A selection of image from a recent shoot:
30 Compressed ARW files: 10:22, 20.73s per image (2.40 Mpix/s)
I also shot some Uncompressed RAW, just to see if there was a difference…
30 Uncompressed ARW files: 9:40, 19.1s per image (2.60 Mpix/s)
And some Lossless Compressed shots for completeness…
30 Lossless Compressed ARW files: 10:18, 20.6s per image (2.41 Mpix/s)
Pretty close across all three. It looks like some of the performance difference is decompressing the original images before processing. All three sets of images have different exposure, ISO, lighting, etc which could affect the results depending on how the dxo algorithms work. I don’t use PhotoLab so have no data from that app.
• 2021 MacBook Pro M1 Max, 32GB RAM, 4TB Internal SSD
• Deep Prime Acceleration Preference: Apple M1 Max
• Method: DeepPRIME
• Optical Corrections: Global lens sharpening and Lens distortion correction ON
• Output format: DNG
• Destination Folder: ‘DxO’ Folder in the original images(s) folder
Not sure who owns the “the DeepPRIME benchmark spreadsheet” Google Sheet linked above, but maybe add a PureRaw tab?